74. The Impact of Different Hinge and Transmembrane Components on the Function of a Novel Fully-Human Anti-CD19 Chimeric Antigen Receptor

Abstract

Listen

Translate article

T cells expressing anti-CD19 chimeric antigen receptors (CARs) have significant activity against B-cell malignancies in humans, but increased efficacy and decreased toxicity of anti-CD19 CAR T cells are needed. Anti-CD19 CARs that have been clinically tested have single-chain variable fragments (scFvs) derived from murine antibodies or in 1 case a humanized murine antibody. Evidence of human anti-mouse immune responses against CARs with murine scFvs has been reported (Jensen et al. Biology of Blood and Marrow Transplantation 2010; Lee et al. Lancet, 2015), providing a rationale for development of less immunogenic CARs. A CAR with a scfv derived from a fully-human antibody would eliminate these human anti-mouse immune responses and perhaps increase the persistence of anti-CD19 CAR T cells. A non-immunogenic CAR might be particularly important if multiple, temporally-separated doses of CAR T cells are administered to the same patient. We have designed and constructed fully-human anti-CD19 CARs encoded by a lentiviral vector. These CARs specifically recognize CD19, and T cells expressing the CARs exhibit a full range of functions including degranulation, cytotoxicity, proliferation, and release of a variety of cytokines. The extracellular scFv component of a CAR is connected to intracellular domains by an extracellular hinge region and a transmembrane (TM) region. In comparing anti-CD19 CARs, we noticed a difference in cytokine production by CARs with hinge and TM regions from the human CD8α molecule compared to CARs with hinge and TM regions from the human CD28 molecule. We designated a fully-human CAR with CD8α hinge and TM domains hu19-CD828Z, and we designated a CAR that was identical to hu19-CD828Z except that it had CD28 hinge and TM domains hu19-28Z. Compared to T cells expressing hu19-CD828Z, T cells expressing hu19-28Z produced much higher levels of interferon gamma (IFNγ), tumor necrosis factor (TNF), and interleukin-2. For example, after an overnight culture with CD19+ NALM6 cells, T cells expressing hu19-CD828Z yielded a mean of 5688 pg/mL IFNγ in the culture supernatant while T cells expressing hu19-28Z yielded 19396 pg/mL IFNγ (P=0.03). Similarly, after an overnight culture with CD19+ NALM6 cells, T cells expressing hu19-CD828Z yielded a mean of 2033 pg/mL TNF in the culture supernatant while T cells expressing hu19-28Z yielded 5008 pg/mL TNF (P=0.007). Cytokine levels were normalized for the percentage of T cells expressing each CAR in each experiment. This same pattern of hinge and TM domains affecting cytokine production was observed for CARs incorporating a different anti-CD19 scFv. In contrast to cytokine production, significant differences in other T-cell functions including degranulation, cytotoxicity, and proliferation were not found when hu19-CD828Z and hu19-28Z were compared. When T cells expressing hu19-CD828Z and hu19-28Z were assessed in a murine lymphoma model, there was not a statistically-significant difference in tumor eradication or survival. We are currently assessing other measures of T-cell activation to obtain a more mechanistic understanding of the differences in T-cell activation with different hinge and TM domains. Most of the clinical toxicity of anti-CD19 CAR T cells is caused by cytokine release. Compared to anti-CD19 CARs with CD28 hinge and TM domains, anti-CD19 CARs with hinge and TM regions from CD8α cause lower levels of cytokine release from T cells in vitro. This decreased cytokine release might reduce cytokine-mediated toxicity in patients. We have initiated a clinical trial of T cells expressing hu19-CD828Z. We will assess cytokine release, immunogenicity, peak blood levels, and long-term persistence of hu19-CD828Z-expressing T cells in humans.